PROC Mixed WARNING: Stopped because of infinite likelihood

[paulusab]

proc mixed data=DIS.THEFINAL noclprint;
	class provfs year wave georegion ARF_code;
	model srrexp = YEAR pctratio georegion ARF_code RNRatio TOTSTAS certdate cream hemom hgm mefavfm nephunkmissm nm / solution ddfm=kr;
	repeated wave / subject=provfs type=ar(1);
	lsmeans year georegion ARF_code/ adjust=tukey;
run;

I'm working with healthcare facility longitudinal data (2013-2016) to predict readmission performance considering a host of variables (captured above in the model statement). 

I've run this model a number of times working my way backwards to significant p-values among variables. After several removals I got this error in the log " WARNING: Stopped because of infinite likelihood". I've read various Q&A's on this forum trying to identify how to move forward and overcome this problem but nothing seems to provide a clear direction of what to do next. It's not duplicate data because that is intended with the structure of this model (facilities with 4 data points over time). Thoughts?

[Reeza]

How many observations do you have?

[paulusab]

21,676